The Immune System & Vaccines

How your body learns a germ by heart

Junior level — plain language, no maths

You are carrying trillions of microbes right now and almost all of them are harmless. Every so often, though, one gets inside that wants to multiply in you, and your body has to deal with it. The first line of defence is blunt and fast: skin, mucus, stomach acid, and roaming cells that swallow anything that looks foreign. They do not know or care which germ it is. They just attack whatever should not be there, and most of the time that is enough.

The clever defence is slower. Scattered through your blood and lymph nodes are cells that each carry one particular shape of receptor - as if you owned a hundred billion keys and one of them happens to fit this germ. When the matching cell finally bumps into it, it starts copying itself furiously, and the copies pour out antibodies: tiny Y-shaped tags that stick to the germ, gum it up and flag it for destruction. Finding that one cell and multiplying it takes about a week. That week is your illness.

When the fight is over most of those cells die off, but some stay behind for years, sometimes for life. They are your memory cells. Meet the same germ a second time and you do not wait a week: the matching cells are already there in large numbers, there is already antibody circulating, and the germ is stopped in a day or two. You never even notice. That is what being immune actually means - not that germs bounce off you, but that you win the race.

A vaccine is that lesson without the illness. It shows your immune system something that looks like the germ but cannot make you sick: a killed version, a weakened one, a single protein from its surface, or just the instructions for making that protein. Your body runs the whole training exercise anyway - finds the matching cells, multiplies them, files them away. And there is a second gift on top. If enough people around you are immune, the germ keeps jumping into people who can already stop it, and the chain of infections snaps. That is what protects the ones who cannot be vaccinated at all: newborns, and people whose immune systems are knocked out by treatment. Slide the vaccinated share up in the simulation and watch the outbreak die.

Antibodies, memory and the herd-immunity threshold

Student level — the core equations

You run two immune systems at once. The innate one is fast, generic and hard-wired: receptors that read patterns common to whole classes of microbe, complement proteins that punch holes, phagocytes that eat. It works in minutes, and on its own it has essentially no memory. The adaptive one is slow, exquisitely specific and it does remember. B cells make antibodies; T cells kill infected cells and give B cells the licence they need to respond at all.

The specificity comes from a trick that looks backwards. Each B cell builds its receptor before it ever meets an antigen, by cutting and splicing a few hundred gene segments into one combination - so the repertoire is generated blindly, in advance, and the germ merely selects the cell that already fits. This is clonal selection. Once selected, the cell divides, and in the germinal centre of a lymph node its antibody genes are deliberately mutated and the best binders picked out again and again: Darwinian evolution, running inside you over days, and it typically improves binding strength by ten to a hundred times.

That process takes time, and the timing is the whole story. A primary response gives detectable antibody at about day 5 to 7 and peaks around day 10 to 14 - which is why a first infection makes you ill for the better part of a week. A secondary response is a different animal: the matching memory cells are ten to a hundred times more numerous, their receptors bind harder, and they need far less antigen to switch on. Antibody appears within one to three days, and the pathogen never gets to the load that makes you feel anything. A vaccine simply hands you the primary response for free, using something that cannot cause the disease.

Now scale up from one body to a town. A pathogen is described by its basic reproduction number \(R_0\): the average number of new infections one case produces in a fully susceptible population. What actually governs an outbreak is the effective number \(R_e = R_0\,s\), where \(s\) is the fraction still susceptible - and an epidemic grows only while \(R_e > 1\). Vaccinate a fraction \(p\) and \(s\) falls to \(1-p\), so the chains break as soon as \(p > 1 - 1/R_0\). That is the herd-immunity threshold, and it is arithmetic, not opinion. Measles, with \(R_0\) between 12 and 18, demands 92 to 95 per cent - which is exactly why measles is the first disease to come back whenever coverage slips.

Clonal selection, affinity maturation and the epidemiology of a threshold

Scholar level — full mathematical depth

01Two systems, and why neither works alone

Innate immunity is germline-encoded: pattern-recognition receptors read conserved microbial signatures that a pathogen cannot easily discard - lipopolysaccharide, flagellin, double-stranded RNA. It responds in minutes and it is the same in you as in your grandmother. Adaptive immunity is built fresh in every individual and takes days. The crucial point is that the second depends on the first. A dendritic cell that picks up antigen only becomes able to prime a naive T cell if it has also received an innate danger signal; antigen without that second signal induces anergy rather than a response. That is a large part of why you do not attack your own proteins - and precisely why vaccines made of purified protein need an adjuvant to work at all.

02A repertoire generated before the question is asked

Developing B and T cells assemble their receptor genes by V(D)J recombination, cutting and rejoining gene segments with RAG1 and RAG2 and adding random nucleotides at the joints. Combinatorial and junctional diversity together give a theoretical repertoire above \(10^{13}\) specificities from a few hundred genes. Macfarlane Burnet's clonal selection theory (1957) drew the right conclusion: the antigen does not instruct the cell what to make, it merely selects a cell that already makes it. Clones that bind self are deleted or silenced during development in bone marrow and thymus - the repertoire is generated blindly and then censored.

03Darwin in a lymph node

A selected B cell enters a germinal centre, where the enzyme AID deliberately mutates its antibody variable genes at roughly \(10^{-3}\) per base per division - about a million times the background rate. Mutants cycle between the dark zone, where they divide and mutate, and the light zone, where they compete to capture scarce antigen and win help from T follicular helper cells. Poor binders die; better binders re-enter and mutate again. Over one to two weeks this raises affinity by one to two orders of magnitude. In parallel, class-switch recombination swaps the constant region - IgM to IgG, IgA or IgE - changing what the antibody does without touching what it binds. The outputs are long-lived plasma cells that home to bone-marrow niches and memory B cells that recirculate.

04Why the second encounter is a different disease

Compare the two responses quantitatively and the advantage is not one thing but four, multiplied together: precursor frequency raised ten- to a hundred-fold, receptor affinity raised ten- to a hundred-fold, a lower threshold for activation, and pre-formed antibody already circulating. The lag from exposure to protective titre collapses from five to seven days down to one to three - and since the pathogen is doubling on a timescale of hours, a few days of head start is the difference between a hard-fought infection and nothing at all. Serological follow-up shows how long the record lasts, and it varies enormously by antigen: extrapolated antibody half-lives run to centuries for measles and mumps, but around eleven years for tetanus and only a few for pertussis. That difference, not policy fashion, is what decides which vaccines need boosters.

05From one host to a population

The compartmental picture is the SIR model: \(\dot S = -\beta SI/N\), \(\dot I = \beta SI/N - \gamma I\), \(\dot R = \gamma I\), with \(R_0 = \beta/\gamma\). Infections grow only while \(\dot I > 0\), that is while \(S/N > 1/R_0\), which gives the threshold \(p_c = 1 - 1/R_0\) directly. Two things are worth saying about \(R_0\). First, it is not a property of the pathogen alone - it is the pathogen multiplied by contact structure, density and behaviour, which is why the same virus is quoted with different values in different settings. Second, an epidemic does not politely stop when \(S/N\) reaches \(1/R_0\): at that instant there is still a large infected population in the pipeline, and it overshoots. The final attack rate solves \(1 - Z = e^{-R_0 Z}\), and for \(R_0 = 3\) that is about 94 per cent of a susceptible population - far more than the 67 per cent that herd immunity would have required.

06Where the clean formula breaks

No vaccine is perfect, so the real requirement is \(p_c = (1 - 1/R_0)/\mathrm{VE}\) - and when \(\mathrm{VE} < 1 - 1/R_0\) the threshold cannot be reached at any coverage. Immunity wanes, at rates that differ by antigen. Antigenic drift moves the target: an influenza threshold computed today is obsolete next season, which is why the vaccine is reformulated annually. And homogeneous mixing is the weakest assumption of all - clustering matters enormously, because a community with 60 per cent coverage inside a country at 95 per cent still sustains outbreaks. That is precisely how measles keeps returning to countries whose national figures look safe. Measles adds a further insult: infection destroys existing memory B cells, erasing an estimated 11 to 73 per cent of a child's accumulated antibody repertoire, so it undoes immunity to diseases already beaten. Eradication, finally, needs three things at once - no animal reservoir, an effective vaccine and reliable diagnosis. Smallpox had all three. Polio nearly does: wild type 2 was declared eradicated in 2015 and type 3 in 2019, leaving type 1 in a handful of districts.